|Publication number||US7139697 B2|
|Application number||US 10/108,661|
|Publication date||Nov 21, 2006|
|Filing date||Mar 27, 2002|
|Priority date||Mar 28, 2001|
|Also published as||EP1246075A2, EP1246075A3, US20020184003|
|Publication number||10108661, 108661, US 7139697 B2, US 7139697B2, US-B2-7139697, US7139697 B2, US7139697B2|
|Inventors||Juha Häkkinen, Markku Mettälä|
|Original Assignee||Nokia Mobile Phones Limited|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (8), Non-Patent Citations (6), Referenced by (71), Classifications (11), Legal Events (6)|
|External Links: USPTO, USPTO Assignment, Espacenet|
The invention relates to determining the language for a character sequence fed into a data processing device.
Various speech recognition applications have been developed in recent years e.g. for vehicle interfaces and mobile stations. Methods are known for mobile stations for calling a desired person by uttering the name of the person into the microphone of a mobile station and establishing a call to the number according to the name uttered by the user. The current methods, however, require the pronunciation of each name to be taught to a telephone or a system in the network. Speaker-independent speech recognition improves the usability of a speech-controlled interface since this training stage is omitted. In e.g. a speaker-independent name selection, the system carries out a text-to-phoneme conversion on the names in a telephone book and compares the name uttered by the user to a determined phoneme sequence. The problem with this method is that the language used by the user in connection with each name is not known. The phoneme sequence produced from the name can thus be erroneous, which means that the identification accuracy is considerably impaired. It is possible for the user to determine the language of the name while entering the name, but as far as usability is concerned this is not a good solution.
Publications U.S. Pat. No. 5,062,143 and EP 1 014276 describe language identification. They disclose methods of identifying a language from a body of text by using “N-grams” (N-letter combinations) or on the basis of occurrence probabilities of short words. In publication U.S. Pat. No. 5,062,143, for example, the most common trigrams (three-letter sequences, such as “abc”) are estimated in each target language from a training text database. In the decoding stage, a language is assigned to a text block if a certain percentage of the trigrams separated from the text is found in a trigram table. The language for which the percentage of matches is greatest is chosen. It is also possible to use common short words, such as determinants, conjunctions and prepositions in each language.
The problem with the prior art solutions is that the N-grams are not very suitable for determining the language of short words, such as names. N-grams require a lot of storage capacity, although different solutions for decreasing the amount of necessary storage capacity do exist. If name recognition is to be carried out in a mobile station, common words (determinants, conjunctions and prepositions) are not available either. Compared with other words, proper names typically follow the common regularities of a language more loosely, which further impairs the operation of the N-gram based methods.
An object of the invention is thus to provide a novel and efficient method for determining a language, and an apparatus implementing the method so that the above-mentioned problems can be avoided. The objects of the invention are achieved by methods, data processing device and a computer program products which are characterized by what is disclosed in the independent claims. Preferred embodiments of the invention are disclosed in the dependent claims.
An object of the invention is to enable the language for a fed character sequence to be determined by means of decision trees. According to an aspect of the invention, character-specific decision trees are then stored in a data processing device, the decision trees describing a probability of at least one language on the basis of the character environment of each character. The environment of a character may consist of adjacent characters to the character or characters located at a greater distance from the character. When the language for a fed character sequence is to be determined, according to an aspect of the invention, decision trees for at least some of the characters of the character sequence are traversed. Each examined character is thus provided with a probability of at least one language. The language for the character sequence is selected on the basis of the language probabilities.
It is to be noted that a probability in a decision tree may be determined to be one, which means that a character is provided with only one language. A character sequence refers to any combination of characters. Typically, a character sequence consists of letters of the alphabet, but it may also comprise punctuation marks (such as commas or spaces). A character sequence may also comprise graph-like characters mainly used in Asian languages. According to an aspect of the invention, a method is provided for forming a decision tree, wherein lexicons to be used of at least two languages are selected that comprise character sequences and language tags associated therewith. The lexicons are combined into one training lexicon, decision trees being formed for different characters to be used for selecting the language for a character sequence by carrying out the following steps of:
According to some further aspects of the invention, computer program products implementing the methods disclosed above are provided.
An advantage of the arrangement of the invention is improved efficiency of language identification. A specific advantage is a very compact presentation format, which enables the arrangement to be readily used also in devices having limited storage capacity, such as mobile stations. The method of the invention has been preliminarily tested and compared with the N-gram based methods, and the method of the invention has given better results particularly when using a relatively closed lexicon. The method has been found efficient particularly in the identification of the language of names, when the lexicon comprises proper names.
According to a preferred embodiment of the invention, the data processing device is a mobile station in a mobile communication system. The present method can then be used for selecting the language for a character sequence fed into the contact data of a mobile station supporting multilingual speech recognition, the decision trees then being formed on the basis of lexicons comprising substantially proper names. A phoneme sequence is formed for the character sequence according to the language selected therefor in accordance with the method. When the mobile station receives speech information from the user, a phoneme sequence substantially according to the speech information supplied by the user is selected, and, furthermore, a character sequence according to said phoneme sequence is selected from the contact data. The advantage of this embodiment is that language selection can be provided for speech recognition using the limited memory resources of the mobile station in an efficient manner.
The invention is now described in closer detail in connection with the preferred embodiments and with reference to the accompanying drawings, in which
The invention can be applied to any data processing device that can be arranged to determine a language for a character sequence fed into the device.
The TE can be any system comprising the above-described equipment and understanding multilingual commands. The TE can be e.g. a personal computer PC or a personal digital assistant PDA device. According to an embodiment, the TE is a mobile station further comprising a mobile station functionality for arranging wireless data transmission with a mobile telephone network. The TE may support any mobile communication standard known to one skilled in the art, e.g. a second generation global system for mobile communication GSM standard, a personal digital cellular PDC standard or a third generation mobile communication standard, such as a system called a universal mobile telecommunication systems UMTS. The TE may also comprise a functionality for accessing a wireless local area network WLAN or a private network, such as a terrestrial trunked radio TETRA.
According to an aspect of the invention, the language for a character sequence fed into the data processing device TE is identified by means of decision trees. The decision trees describe language probabilities on the basis of the environment of the characters, i.e. the context.
The nodes RN, IN comprise attributes Attr., which are questions about the characters surrounding the characters, e.g. “Is Y the first character to the right of the character?”. According to an embodiment of the invention, the questions may concern the characters on either side of the character, enabling the environment of the character to be mapped as comprehensively as possible. Since only information on the environment is used while no frequency information on different languages is stored in the environment corresponding to the particular node, an extremely compact presentation format is achieved. By answering the questions, the nodes are examined typically until a leaf is met. In the example of
It is to be noted that the decision trees for different characters can be formed 301 in a separate training device which is not the TE. Typically, decision trees are trained once, e.g. application specifically, and stored in the memory of the training device, wherefrom the decision trees can be transferred to the data processing devices TE through external memory means (e.g. CD-ROM) or a network. The decision trees can be stored 302 in the memory MEM of different data processing devices e.g. during the manufacturing of the TE or in connection with storing an application in need of language selection. The language selector block LE of the TE thus carries out steps 304 to 308 and the training device carries out step 301 (and also 302), which is described in closer detail in
When a character sequence is fed 303 into the data processing device, its language selector block LE retrieves 304 a decision tree for a character of the character sequence from the memory MEM and traverses 305 the decision tree. The characters of the character sequence do not have to be examined in a particular order but characters whose decision trees will be traversed can be selected first. After the decision tree has been traversed (a leaf has been met or there is no way to proceed from a node), a probability of at least one language is achieved for the character according to the value of the node or the leaf. Preferably, all characters of the character sequence are examined, which means returning from step 306 to step 304 if the decision trees for all characters have not been examined. The LE can naturally retrieve 304 the decision trees for all characters from the memory MEM in one go, which means that the decision trees for each character do not have to be retrieved separately from the memory. In the decision tree, the probability can also be just zero or one, in which case the decision tree may give the characters only one language. When all characters of the character sequence have been provided with language probabilities from the decision trees, the language selector block LE automatically selects 307 the language on the basis of said probabilities.
According to an embodiment, the language is selected 307 on the basis of adding up the language probabilities of different characters. The language for which the total sum of probabilities is greatest is selected as the language of the character sequence.
According to another embodiment, the language which is most probable for most characters of the character sequence is selected 307 as the language of the character sequence. If the same number of characters have two (or more) languages as the most probable language, the language e.g. for which the probabilities are greatest can be selected. All the most probable languages of the characters are then treated equally.
According to an embodiment of the invention alternative to step 307, the user of the data processing device TE is provided 308 with the most probable languages to select from. The user interface of the TE can be used for showing first the language which, on the basis of the decision trees, is the most probable one. Preferably, at least one alternative language is also shown, which is the second most probable language. The user can directly accept the suggested language or select another language for the character sequence. The TE waits until the user has selected the language and selects 308 the language selected by the user to eventually be the language of the character sequence. The advantage of this embodiment is that even fewer errors are made in the language selection when the language of the character sequence is confirmed by the user.
Usually the data processing devices TE are provided with a default language, e.g. the language selected for the user interface. According to an embodiment, in dubious situations (when many substantially equally probable languages exist), the user is provided 308 with the default language to select or the default language is automatically selected 307. According to still another embodiment, more than one language can be selected 307, 308 for the character sequence. This may be necessary when the character sequence exists in many languages, e.g. the name “Peter” gives English and German as the languages.
In the following, the way in which a decision tree is trained for a character will be described in closer detail (step 301 in
Fin, Fin, Fin, Fin
Eng, Eng, Eng, Eng, Eng
Ger, Ger, Ger, Ger, Ger
Eng, Eng, Eng
Ger, Ger, Ger, Ger
When the decision tree is trained for a given character, all training cases are taken into account. A training case of a character comprises the environment of the character and a corresponding language tag in the training lexicon. In the course of the training, the decision tree is grown and the nodes are divided into sub-nodes on the basis of an information-theoretic optimization criterion. A root node RN is first divided at least into two internal nodes IN. In order to be able to divide a node RN, IN, an attribute, i.e. a question concerning the environment of the character, has to be selected. Different attributes are formed 403 and compared 404 with each other. The one of the attributes which maximizes the optimization criterion is selected 405 as a new node. Information gain is used as an optimization criterion. In order to be able to calculate the information gain obtained by the splitting, the language distribution before the splitting must be known. Entropy E can be calculated on the basis of the language tag distribution in the node in the following manner:
wherein fi is the relative frequency of language tag occurrence (in the words in the training lexicon that match the particular node), and N is the number of language tags. After entropy division, an average entropy Es is calculated for the attribute from the entropies of subsets. If Ej s describes the entropy of subset j after the division, the average entropy after the division is
where |S| is the total number of training cases in the root node RN, |Sj| is the number of training cases in the jth subset and K is the number of subsets. The information gain for the attribute is obtained from
Information gain is calculated 403 for each attribute, and the attribute which has the largest information gain is selected 405. The nodes are divided 403 to 405 as long as the information gain is larger than zero and the entropies of the nodes can be improved by dividing the nodes. In addition to meeting the information gain criterion, the node IN, RN can be divided only if at least two internal nodes exist that have at least the minimum number of training sequences after the division.
According to an embodiment of the invention, the nodes are divided (406→403) until a predetermined ending criterion is met on the basis of a checking procedure 406. An ending criterion may consist of e.g. restricting the decision tree to be smaller than a predetermined size, restricting the decision tree to be of a given depth (how many steps can at most be formed after the root node RN in order to reach the most distant leaf), or the fact that the information gain obtained by the splitting remains below a predetermined limit value. Naturally, a combination of the above-mentioned criteria can be used. It is also feasible to restrict the size of a decision tree by adjusting the ending criteria or by applying a pruning method. When no nodes are added any longer, leaves L are added 407 to the decision tree. The leaves L and preferably also the nodes RN, IN comprise a probability of at least one language. The probability can be one, expressed only in the form of a language tag, or the probabilities of different languages can be calculated in step 407 and stored at least in the leaves L. Next, the trained decision tree can be stored 408 in a memory device, such as in the internal memory MEM of the data processing device TE. The decision trees can be stored 302 in the memory MEM of the data processing device TE from external memory means or by transferring them from the network e.g. over the Internet. Some characters may have an extremely simple decision tree, e.g. the decision tree of the character “ü” can explicitly indicate that the language is German.
According to an embodiment, in addition to the decision tree, the data processing device TE is also provided with determined rules to overrule or complete the result of a decision tree. For example, German can be determined as the language of a character sequence if the sequence comprises the character “ü”, which means that the there is no need to traverse the decision trees at all. The rules can be stored 302 in the memory of the TE and they can in special situations be employed to make the language selecting process even faster.
The invention can be used in any application wherein a language is to be determined from a fed character sequence. A typical application in which the language determination of the invention can be used is a multilingual speaker-independent speech recognition application. For this application, phoneme conversions are determined in a mobile station for the names fed into the contact data by the user. A phoneme conversion is typically obtained using a look-up table or an automatic text-to-phoneme mapping. The mapping is typically language-dependent while look-up tables are always language-dependent, i.e. different languages have text-to-phoneme sources of their own. In order for speech recognition to work appropriately, the speech recognition application must thus first determine (steps 304 to 307/308) the languages of the character sequences of the contact data so as to enable a correct phoneme sequence to be attached to the name fed by the user into the contact data.
A phoneme sequence can be determined for the character sequence from a language-specific pronunciation dictionary. The phoneme sequence is attached to the tag of the character sequence, which can be called a name-tag. When the user utters a word, the speech recognition application, by comparing the phoneme sequences, is able to choose a correct character sequence (name) and e.g. establish a call to the number attached to the character sequence. If there are several substantially equally probable languages, the name can also be associated with several languages either automatically 307 or based on a choice 308 made by the user. The phoneme sequences for at least the two most probable languages can then be formed in the mobile station, in which case either one of the pronunciations is acceptable. The invention can naturally also be applied to speech recognition for other types of device or applications. Utilizing the language selection of the invention, users speaking different languages can be provided with a speech-controlled user interface without the user being compelled to program speech commands in advance.
A second application wherein the invention can be utilized is a search for street addresses or names of cities on the basis of speech e.g. in a navigation system. A third application is a database search on the basis of the name of the person executing the search, wherein the language and pronunciation of the name are determined. As a fourth application could be mentioned language determination in a spelling control unit, as a fifth application language determination from a scanned text (e.g. using an optical character recognition OCR technique), and as a sixth application language determination for a user interface on the basis of a name while the user is logging in to a system. A seventh application could be language determination directly from a phoneme sequence formed from the speech of the user e.g. for machine translation and interpretation.
It is obvious to one skilled in the art that as technology advances, the basic idea of the invention can be implemented in many different ways. The invention and its embodiments are thus not restricted to the examples described above but can vary within the scope of the claims.
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US5062143||Feb 23, 1990||Oct 29, 1991||Harris Corporation||Trigram-based method of language identification|
|US5548507 *||Mar 14, 1994||Aug 20, 1996||International Business Machines Corporation||Language identification process using coded language words|
|US5805832 *||Jun 2, 1995||Sep 8, 1998||International Business Machines Corporation||System for parametric text to text language translation|
|US6016471||Apr 29, 1998||Jan 18, 2000||Matsushita Electric Industrial Co., Ltd.||Method and apparatus using decision trees to generate and score multiple pronunciations for a spelled word|
|US6026410 *||Feb 10, 1997||Feb 15, 2000||Actioneer, Inc.||Information organization and collaboration tool for processing notes and action requests in computer systems|
|US6912499 *||Aug 31, 1999||Jun 28, 2005||Nortel Networks Limited||Method and apparatus for training a multilingual speech model set|
|EP1014276A2||Dec 17, 1999||Jun 28, 2000||Xerox Corporation||Automatic language identification using both N-Gram and word information|
|GB2338369A||Title not available|
|1||"A Study Of N-Gram And Decision Tree Letter Language Modeling Methods", Potamianos et al., Speech Communication, vol. 24(3), 1998.|
|2||"A Tree-Based Statistical Language Model For Natural Language Speech Recognition", Bahl et al., IEEE Transactions on Acoustics, Speech and Signal Processing, vol. 37, No. 7, 1989.|
|3||"Automatic Construction of Decision Trees From Data: A Multi-Disciplinary Survey", Murthy, Data Mining and Knwoledge Discovery, vol. 4, 1998.|
|4||"Incorporating POS Tagging Into Language Modeling" Heeman et al., Proceeding of the 5<SUP>th </SUP>European Conference on Speech Communication and Technology, 1997.|
|5||"Text Classification And Segmentation Using Minimum Cross-Entropy", Teahan, International Conference on Content-Based Multimedia Information Access, 2000.|
|6||"Two Decades Of Statistical Language Modeling: Where Do We Go From Here?", Rosenfeld, Proceedings of the IEEE, 2000.|
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US7562007 *||Jun 16, 2004||Jul 14, 2009||Samsung Electronics Co., Ltd.||Method and apparatus for recognizing language input mode and method and apparatus for automatically switching language input modes using the same|
|US7689409 *||Dec 11, 2003||Mar 30, 2010||France Telecom||Text language identification|
|US8185376 *||Mar 20, 2006||May 22, 2012||Microsoft Corporation||Identifying language origin of words|
|US8326602 *||Jun 5, 2009||Dec 4, 2012||Google Inc.||Detecting writing systems and languages|
|US8380507||Mar 9, 2009||Feb 19, 2013||Apple Inc.||Systems and methods for determining the language to use for speech generated by a text to speech engine|
|US8457946 *||Apr 26, 2007||Jun 4, 2013||Microsoft Corporation||Recognition architecture for generating Asian characters|
|US8468011 *||Jun 5, 2009||Jun 18, 2013||Google Inc.||Detecting writing systems and languages|
|US8583418 *||Sep 29, 2008||Nov 12, 2013||Apple Inc.||Systems and methods of detecting language and natural language strings for text to speech synthesis|
|US8635061 *||Oct 14, 2010||Jan 21, 2014||Microsoft Corporation||Language identification in multilingual text|
|US8712776||Sep 29, 2008||Apr 29, 2014||Apple Inc.||Systems and methods for selective text to speech synthesis|
|US8751238||Feb 15, 2013||Jun 10, 2014||Apple Inc.||Systems and methods for determining the language to use for speech generated by a text to speech engine|
|US8892446||Dec 21, 2012||Nov 18, 2014||Apple Inc.||Service orchestration for intelligent automated assistant|
|US8903716||Dec 21, 2012||Dec 2, 2014||Apple Inc.||Personalized vocabulary for digital assistant|
|US8930191||Mar 4, 2013||Jan 6, 2015||Apple Inc.||Paraphrasing of user requests and results by automated digital assistant|
|US8942986||Dec 21, 2012||Jan 27, 2015||Apple Inc.||Determining user intent based on ontologies of domains|
|US9117447||Dec 21, 2012||Aug 25, 2015||Apple Inc.||Using event alert text as input to an automated assistant|
|US9262612||Mar 21, 2011||Feb 16, 2016||Apple Inc.||Device access using voice authentication|
|US9300784||Jun 13, 2014||Mar 29, 2016||Apple Inc.||System and method for emergency calls initiated by voice command|
|US9318108||Jan 10, 2011||Apr 19, 2016||Apple Inc.||Intelligent automated assistant|
|US9330720||Apr 2, 2008||May 3, 2016||Apple Inc.||Methods and apparatus for altering audio output signals|
|US9338493||Sep 26, 2014||May 10, 2016||Apple Inc.||Intelligent automated assistant for TV user interactions|
|US9368114||Mar 6, 2014||Jun 14, 2016||Apple Inc.||Context-sensitive handling of interruptions|
|US9430463||Sep 30, 2014||Aug 30, 2016||Apple Inc.||Exemplar-based natural language processing|
|US9483461||Mar 6, 2012||Nov 1, 2016||Apple Inc.||Handling speech synthesis of content for multiple languages|
|US9495129||Mar 12, 2013||Nov 15, 2016||Apple Inc.||Device, method, and user interface for voice-activated navigation and browsing of a document|
|US9502031||Sep 23, 2014||Nov 22, 2016||Apple Inc.||Method for supporting dynamic grammars in WFST-based ASR|
|US9535895 *||Mar 17, 2011||Jan 3, 2017||Amazon Technologies, Inc.||n-Gram-based language prediction|
|US9535906||Jun 17, 2015||Jan 3, 2017||Apple Inc.||Mobile device having human language translation capability with positional feedback|
|US9548050||Jun 9, 2012||Jan 17, 2017||Apple Inc.||Intelligent automated assistant|
|US9576574||Sep 9, 2013||Feb 21, 2017||Apple Inc.||Context-sensitive handling of interruptions by intelligent digital assistant|
|US9582608||Jun 6, 2014||Feb 28, 2017||Apple Inc.||Unified ranking with entropy-weighted information for phrase-based semantic auto-completion|
|US9589564 *||Feb 5, 2014||Mar 7, 2017||Google Inc.||Multiple speech locale-specific hotword classifiers for selection of a speech locale|
|US9606986||Sep 30, 2014||Mar 28, 2017||Apple Inc.||Integrated word N-gram and class M-gram language models|
|US9620104||Jun 6, 2014||Apr 11, 2017||Apple Inc.||System and method for user-specified pronunciation of words for speech synthesis and recognition|
|US9620105||Sep 29, 2014||Apr 11, 2017||Apple Inc.||Analyzing audio input for efficient speech and music recognition|
|US9626955||Apr 4, 2016||Apr 18, 2017||Apple Inc.||Intelligent text-to-speech conversion|
|US9633004||Sep 29, 2014||Apr 25, 2017||Apple Inc.||Better resolution when referencing to concepts|
|US9633660||Nov 13, 2015||Apr 25, 2017||Apple Inc.||User profiling for voice input processing|
|US9633674||Jun 5, 2014||Apr 25, 2017||Apple Inc.||System and method for detecting errors in interactions with a voice-based digital assistant|
|US9646609||Aug 25, 2015||May 9, 2017||Apple Inc.||Caching apparatus for serving phonetic pronunciations|
|US9646614||Dec 21, 2015||May 9, 2017||Apple Inc.||Fast, language-independent method for user authentication by voice|
|US9658989 *||Aug 25, 2008||May 23, 2017||International Business Machines Corporation||Apparatus and method for extracting and manipulating the reading order of text to prepare a display document for analysis|
|US9668024||Mar 30, 2016||May 30, 2017||Apple Inc.||Intelligent automated assistant for TV user interactions|
|US9668121||Aug 25, 2015||May 30, 2017||Apple Inc.||Social reminders|
|US9697820||Dec 7, 2015||Jul 4, 2017||Apple Inc.||Unit-selection text-to-speech synthesis using concatenation-sensitive neural networks|
|US9697822||Apr 28, 2014||Jul 4, 2017||Apple Inc.||System and method for updating an adaptive speech recognition model|
|US9711141||Dec 12, 2014||Jul 18, 2017||Apple Inc.||Disambiguating heteronyms in speech synthesis|
|US9715875||Sep 30, 2014||Jul 25, 2017||Apple Inc.||Reducing the need for manual start/end-pointing and trigger phrases|
|US9720903 *||Jul 22, 2013||Aug 1, 2017||Robert D. New||Method for parsing natural language text with simple links|
|US9721566||Aug 31, 2015||Aug 1, 2017||Apple Inc.||Competing devices responding to voice triggers|
|US9734193||Sep 18, 2014||Aug 15, 2017||Apple Inc.||Determining domain salience ranking from ambiguous words in natural speech|
|US9760559||May 22, 2015||Sep 12, 2017||Apple Inc.||Predictive text input|
|US9785630||May 28, 2015||Oct 10, 2017||Apple Inc.||Text prediction using combined word N-gram and unigram language models|
|US9798393||Feb 25, 2015||Oct 24, 2017||Apple Inc.||Text correction processing|
|US9818400||Aug 28, 2015||Nov 14, 2017||Apple Inc.||Method and apparatus for discovering trending terms in speech requests|
|US20040138869 *||Dec 11, 2003||Jul 15, 2004||Johannes Heinecke||Text language identification|
|US20040260536 *||Jun 16, 2004||Dec 23, 2004||Samsung Electronics Co., Ltd.||Method and apparatus for recognizing language input mode and method and apparatus for automatically switching language input modes using the same|
|US20050197837 *||Mar 8, 2004||Sep 8, 2005||Janne Suontausta||Enhanced multilingual speech recognition system|
|US20060100854 *||Oct 11, 2005||May 11, 2006||France Telecom||Computer generation of concept sequence correction rules|
|US20070219777 *||Mar 20, 2006||Sep 20, 2007||Microsoft Corporation||Identifying language origin of words|
|US20080126093 *||Nov 28, 2006||May 29, 2008||Nokia Corporation||Method, Apparatus and Computer Program Product for Providing a Language Based Interactive Multimedia System|
|US20080270118 *||Apr 26, 2007||Oct 30, 2008||Microsoft Corporation||Recognition architecture for generating Asian characters|
|US20090063965 *||Aug 25, 2008||Mar 5, 2009||International Business Machines Corporation||Preparing a display document for analysis|
|US20100082329 *||Sep 29, 2008||Apr 1, 2010||Apple Inc.||Systems and methods of detecting language and natural language strings for text to speech synthesis|
|US20100312545 *||Jun 5, 2009||Dec 9, 2010||Google Inc.||Detecting Writing Systems and Languages|
|US20120095748 *||Oct 14, 2010||Apr 19, 2012||Microsoft Corporation||Language Identification in Multilingual Text|
|US20120239379 *||Mar 17, 2011||Sep 20, 2012||Eugene Gershnik||n-Gram-Based Language Prediction|
|US20140019122 *||Jul 22, 2013||Jan 16, 2014||Robert D. New||Method for Parsing Natural Language Text|
|US20140108002 *||Oct 17, 2012||Apr 17, 2014||Research In Motion Limited||Method and system for updating display information based on detected language of a received message|
|US20150221305 *||Feb 5, 2014||Aug 6, 2015||Google Inc.||Multiple speech locale-specific hotword classifiers for selection of a speech locale|
|US20150309996 *||Apr 9, 2015||Oct 29, 2015||Xiaofeng Han||Information processing system, information processing apparatus, and information processing method|
|U.S. Classification||704/9, 704/8|
|International Classification||G06F17/20, G06F17/28, G06F17/27|
|Cooperative Classification||G06F17/2715, G06F17/275, G06F17/2775|
|European Classification||G06F17/27R4, G06F17/27L, G06F17/27A4|
|May 31, 2002||AS||Assignment|
Owner name: NOKIA CORPORATION, FINLAND
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:HAKKINEN, JUHA;METTALA, MARKKU;REEL/FRAME:012935/0675;SIGNING DATES FROM 20020425 TO 20020429
|May 3, 2010||FPAY||Fee payment|
Year of fee payment: 4
|Apr 23, 2014||FPAY||Fee payment|
Year of fee payment: 8
|May 9, 2015||AS||Assignment|
Owner name: NOKIA TECHNOLOGIES OY, FINLAND
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:NOKIA CORPORATION;REEL/FRAME:035601/0901
Effective date: 20150116
|Sep 21, 2017||AS||Assignment|
Owner name: OMEGA CREDIT OPPORTUNITIES MASTER FUND, LP, NEW YO
Free format text: SECURITY INTEREST;ASSIGNOR:WSOU INVESTMENTS, LLC;REEL/FRAME:043966/0574
Effective date: 20170822
|Sep 25, 2017||AS||Assignment|
Owner name: WSOU INVESTMENTS, LLC, CALIFORNIA
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:NOKIA TECHNOLOGIES OY;REEL/FRAME:043953/0822
Effective date: 20170722